LGMar 8, 2018
Some Approximation Bounds for Deep Networks
arXiv:1803.02956v1
Originality Synthesis-oriented
AI Analysis
This work addresses theoretical understanding of deep learning models, but appears incremental as it builds on existing bounds and architectures.
The paper introduces new approximation bounds for deep networks and proposes novel architectures, providing theoretical insights into the success of autoencoders and ResNets.
In this paper we introduce new bounds on the approximation of functions in deep networks and in doing so introduce some new deep network architectures for function approximation. These results give some theoretical insight into the success of autoencoders and ResNets.